Alterations of Enteric Microbiota in Patients with a Normal Ileal Pouch Are Predictive of Pouchitis

Abstract

Objective:

To examine whether patients with a mature normal pouch [> 1 year post ileostomy closure] have microbial stool characteristics that can predict pouch inflammation.

Design:

Patients undergoing pouch surgery were recruited prospectively. Microbiota analysis of faecal samples was by 16S rRNA gene pyrosequencing. All patients had a normal pouch at baseline [T1]. Those without pouchitis during the first year of follow-up [T2] comprised the ‘Normal Pouch-sustained’ group and those who had experienced an episode of pouchitis comprised the ‘Pre-Pouchitis’ group.

Results:

Twenty patients were recruited (age 53.6±13.1 years, pouch age [time from ileostomy closure] 8.1±5.1 years). Seven patients developed pouchitis during follow-up [within 265±93.6 days] and they were assigned to the Pre-Pouchitis group at T1: they had a decreased microbial diversity at T1 compared with the Normal Pouch-sustained patients [n = 13]. The Shannon diversity index for the Pre-Pouchitis patients was 3.4 vs 4.23 for the Normal Pouch-sustained patients [p = 0.011]. There were no substantial group differences in high taxonomic levels [order or above]. The genus Ruminococcus was significantly decreased in the Pre-Pouchitis patients’ samples compared with those of the Normal Pouch-sustained patients (0.19% vs 0.78%, respectively, false discovery rate [FDR] = 0.05). The linear discriminant analysis with effect size estimation algorithm revealed that Lachnospira and Coprococcus genera were also decreased among Pre-Pouchitis patients compared with Normal Pouch-sustained patients [0.6% vs 1.95% and 2.1% vs 4%, respectively].

Conclusions:

Patients with a normal mature pouch may be predisposed to acute pouchitis when faecal microbial diversity and certain microbial groups are decreased. These findings may aid in risk stratification of those patients.

1. Significance of this study

1.1. What is already known on this subject?

• Ileal pouch-anal anastomosis is the surgical procedure of choice for patients with ulcerative colitis who suffer from refractory disease or dyplasia/cancer.

• Up to 60% of pouch patients develop pouchitis.

• A decrease in enteric microbial diversity and changes in microbial composition have been associated with pouchitis.

1.2. What are the new findings?

• Decreased faecal microbial diversity in ileal pouch patients may be predictive of pouch inflammation up to 1 year before inflammation occurs.

• Several genera, such as Ruminococcos [Ruminococcaceae family], Coprococcus and Lachnospira, both belonging to the Lachnospiraceae family, are increased in faecal samples of patients who do not develop pouchitis.

• Faecal microbial diversity is decreased during the first year after surgery and gradually increases in the years following pouch surgery.

1.3. How might it impact on clinical practice in the foreseeable future?

• These findings may provide an important tool for the clinician to predict which patients may be at risk to develop pouch inflammation.

• Treatments aimed at increasing microbial diversity may help and should be tested for preventing pouchitis.

2. Introduction

The incidence of ulcerative colitis has been increasing globally during the past few decades.¹ Despite the introduction of new and more effective medications, approximately 30% of ulcerative colitis patients may undergo surgery, mostly due to refractory disease or dysplasia/colon cancer.² The surgery of choice is restorative proctocolectomy and ileal pouch anal-anastomosis [IPAA], in which the colon and rectum are resected and a reservoir [pouch] is created from the unaffected small bowel.^3–6 Unfortunately, up to 60% of those patients may develop inflammation of the pouch [pouchitis], usually within the first year post ileostomy closure.^7,8

The pathogenesis of pouchitis and of inflammatory bowel disease including ulcerative colitis is presumed to involve abnormal immune responses to the commensal microbiota in genetically predisposed individuals.^7,9 Recent evidence has suggested that a decrease in enteric microbial diversity, the loss of beneficial bacteria and an increase in bacterial groups associated with inflammation contribute to pouchitis.^9–15 However, it is still unclear whether alterations in microbial composition are the cause or the result of inflammation.

We considered that longitudinal follow-up of patients who have a pouch may provide a source for detecting microbial alterations that had occurred and that may drive pouch inflammation. We further assumed that such an investigation may also add insight into the pathogenesis of the small intestinal inflammation occurring in some patients with Crohn’s disease.

Here, we aimed to explore whether microbial alterations are predictive of pouch inflammation. We prospectively followed up patients with a normal pouch until they developed pouchitis and compared them with patients who did not. We were able to demonstrate that decreased enteric microbial diversity predicted the onset of pouch inflammation within 1 year.

3. Materials and Methods

3.1. Study patients

Patients diagnosed as having ulcerative colitis, who underwent pouch surgery, were prospectively recruited at the Comprehensive Pouch Clinic, the Tel Aviv Sourasky Medical Center, a tertiary referral centre for inflammatory bowel disease [IBD] and the national referral centre for pouch patients. Both an IBD-oriented gastroenterologist [ID] and a colorectal surgeon [HT] routinely examined all patients.

The inclusion criteria were: patients who had a pouch due to ulcerative colitis and who; were at least 1 year post ileostomy closure; had a clinical definition of a normal pouch or Pouchitis Disease Activity Index [PDAI] < 7 [see below] at baseline; had an available stool sample for 16S rRNA gene pyrosequencing; and had a follow-up time of > 1 year or until an episode of acute pouchitis was diagnosed. Exclusion criteria were: having received antibiotic treatment within 3 months preceding stool sampling; and a PDAI clinical subscore of ≥ 2.

This study received the Tel Aviv Medical Center’s ethics review board approval [IRB# 0467-10, 15/12/2010] and each participant provided written informed consent.

3.2. Definition of pouch behaviour and disease activity

Disease activity was assessed during the longitudinal follow-up at two time points: time point 1 [T1], the first visit at which patients were clinically assessed, enrolled in the study and stool sample wascollected; and time point 2 [T2], the second visit at which clinical assessment was conducted and clinical outcome determined. Disease activity assessment was based on the PDAI¹⁶ including its clinical, endoscopic and histological subscores. A normal pouch was defined as a PDAI ≤ 6 and no antibiotic therapy during the preceding 3 months. Pouchitis was defined as a PDAI ≥ 7 or if antibiotic therapy had been required as a therapy for presumed pouchitis based on the clinical subscore of the PDAI. Patients were classified into one of two groups: those with a normal pouch at both T1 and T2 [Normal Pouch-sustained group] and those with a normal pouch at T1 who were subsequently diagnosed with pouchitis at T2 within 1 year from T1 [Pre-Pouchitis group] [Figure 1].

Endoscopy was usually performed within 1 month of clinical assessment. In cases when the interval to endoscopy was longer, it was performed only if the clinical symptoms at endoscopy and at T1 remained the same.

3.3. Clinical data and sample collection

Demographics and clinical data were collected during clinic visits. Time from IBD diagnosis to restorative proctocolectomy and ileal pouch anal-anastomosis, as well as pouch age [the time interval between ileostomy closure and sampling] were documented. Faecal samples were collected in sterile cups at T1 and immediately frozen at -80°C until processing for the determination of the faecal calprotectin level and for bacterial genomic DNA extraction.

3.4. Faecal sample processing

Faecal samples were thawed at room temperature and DNA was extracted using the PowerLyzer PowerSoil DNA Isolation Kit [MO-BIO, Carlsbad, CA]. The OMNI Bead Ruptor 24 Homogenizer [OMNI International, Kennesaw, GA] was used for sample homogenization at the following settings: speed 5.65 m/s, cycles 02, run time 0:45min and dwell time 0:30min. Extracted DNA samples were stored at -80°C. The presence of bacterial DNA was validated by polymerase chain reaction, with the following primers: 27F 5′-AGAGTTTGATCMTGGCTCAG-3′, 338R 5′-GCTGCCTCCCGTAGGAGT-3′.

3.5. 16S ribosomal RNA pyrosequencing

The 16S region ranging from position 28 to 519 was sequenced [Research and Testing Laboratory, Lubbock, TX] by the bacterial 16S-based tag-encoded FLX amplicon pyrosequencing method¹⁷ using the primers 28F 5′-GAGTTTGATCNTGGCTCAG-3′, 519R 5′GTNTTACNGCGGCKGCTG-3′. Briefly, sequences that were less than 200bp or greater than 1000bp in length, that contained incorrect primer sequences or that contained more than one ambiguous base were discarded. The average number of sequences per sample was 4671 [range 1899–10313].

3.6. Data and statistical analysis

Sequencing results were analysed using the Quantitative Insights Into Microbial Ecology [QIIME] package.¹⁸ Uclust¹⁹ was used to pick OTUs [o perational taxonomic units]at a 0.97 similarity level, and taxonomy was assigned by BLAST [E value < 0.001] using the Greengenes database of May 2013 in a closed reference manner. Chimeric sequences were identified by ChimeraSlayer and removed. UniFrac analysis was used to calculate the distance between OTUs/bacterial communities on a phylogenetic tree, and principal coordinates were generated using unweighted and weighted UniFrac distances for all samples.^20,21 Weighted and unweighted UniFrac distances represent compositional dissimilarity/heterogeneity [often referred to as beta diversity]. However, since an unweighted analysis considers only the presence or absence of bacterial groups, all bacterial groups will have a similar impact on the UniFrac distances. In a weighed analysis, the relative abundance of the bacterial groups within the communities is accounted for and so the more abundant bacterial groups have a greater impact on the UniFrac distances. Principal coordinates Analysis [PCoA] plots were used to visualize the similarities or dissimilarities that best represent the pair-wise distances between sample groups. Statistical differences in weighted and unweighted UniFrac distances between groups [beta diversity] were tested using analysis of similarity [ANOSIM, implemented in QIIME] by permutation of group membership with 999 replicates. The test statistic R, which measures the strength of the results, ranges from −1 to 1, in which R = 1 signifies that groups are completely separated and R = 0 signifies that samples within a group are as distant from one another as are samples from different groups. The number of observed bacterial species and the Shannon index of diversity [a measure of diversity taking into account richness and evenness of OTUs]²² were compared following rarefaction of OTUs using Student’s t test. The proportions of core bacterial taxa [i.e. taxa that appear in more than 50% of samples] were normalized as previously described¹⁸ and compared using Student’s t test.

Linear discriminant analysis with effect size estimation [LEfSe]²³ was performed on genus-level OTU tables using the online analysis tool available from [http://huttenhower.sph.harvard.edu/galaxy/]. The LEfSe algorithm allows identification of taxa that most strongly differentiate one group from another [i.e. biomarkers]. A p-value < 0.05 was considered significant. For multiple comparisons, a false detection rate [FDR] < 0.1 was considered significant. Student’s t test and graphs were performed using the GraphPad Prism software version 5 for Windows.

4. Results

Twenty patients were recruited and followed up during a period of 40 months. They were all at least 1 year post ileostomy closure and were classified as having a normal pouch at baseline. There were 13 Normal Pouch-sustained patients and 7 patients who developed pouchitis within 1 year of study recruitment [Pre-Pouchitis] [Figure 1A]. The Normal Pouch-sustained patients were further followed up [for up to 4 years] in order to determine if and when they developed pouchitis. Time to pouchitis in the Pre-Pouchitis group was significantly shorter than in the Normal Pouch-sustained group [265±93 vs 846±65 days, respectively, p < 0.0001] [Figure 1B].

There were no significant differences between the two groups in terms of gender, pouch age, aetiology for surgery, initial PDAI score, faecal calprotectin level or probiotic use [Table 1]. However, patients who were classified into the Normal Pouch-sustained group were significantly older than the Pre-Pouchitis patients [59.7±9.6 vs 42.3±11.4 years, respectively, p < 0.002].

4.1. Microbial diversity

Microbial diversity [α-diversity], assessed through rarefaction curves and the Shannon diversity index, was significantly increased in the Normal Pouch-sustained group compared with the Pre-Pouchitis group [p < 0.01] [Figure 2]. This was already apparent at the time of study entry [T1], when there was no evidence of pouch inflammation, as assessed by endoscopy, histology and faecal calprotectin level [Supplementary Figure 1, available as Supplementary data at ECCO-JCC online].

4.2. Microbial communities

Microbial communities were comparable between the two groups of patients. A weighted UniFrac analysis did not reveal any significant differences [Supplementary Figure 2, available as Supplementary data at ECCO-JCC online], and an unweighted UniFrac analysis revealed a modest difference between the two groups [Figure 3A].

According to an LEfSe analysis, three genera that belong to the Clostridiales order were more abundant in the faecal samples of the Normal Pouch-sustained group compared with the Pre-Pouchitis group [Figure 3B]. However, the relative abundance of the taxa at the phyla, class, order and family levels did not reveal any significant differences between the groups when analysed using a Student’s t test adjusted for multiple comparisons [FDR < 0.1] [Supplementary Figure 2, phyla analysis]. At the genus level, two bacteria, Coprococcus and Ruminococcous, were significantly more common in the Normal Pouch-sustained group [p < 0.05 compared with the Pre-Pouchitis group [Figure 3C,D]. After adjusting for multiple comparisons, however, only the Ruminococcus genus remained significantly more common in the Normal Pouch-sustained group [FDR = 0.05].

We further analysed the data to examine whether the differences between the two groups could be attributed to the difference in patients’ ages. An unweighted analysis of bacterial composition did not reveal any differences between the two groups, nor was the Shannon diversity index significantly different between the two groups. Furthermore, any trend towards age-related differences disappeared when each group was analysed separately [Supplementary Figure 3, available as Supplementary data at ECCO-JCC online].

4.3. Microbial diversity was low during the first year post ileal pouch-anal anastomosis surgery

During the first year after pouch surgery, the observed high rate of developing pouch inflammation may be related to an unstable microbiota profile.^7,24,25 Indeed, in the current investigation, microbial diversity was significantly lower in the five Normal Pouch-sustained patients with a pouch age < 1 year compared with the 13 Normal Pouch-sustained patients with a pouch age > 1 year [Figure 4A]. Interestingly, a 3-year follow-up post pouch surgery of these patients demonstrated a steady increase in microbial diversity [n = 4, Figure 4B], becoming more similar to that of the Normal Pouch-sustained patients.

5. Discussion

The results of this prospective study demonstrated that microbial diversity in an ileal pouch may be predictive of pouch inflammation up to 1 year before inflammation becomes apparent clinically. Out of more than 400 patients with ulcerative colitis who underwent pouch surgery and are prospectively followed up at our centre,^8,9 we focused upon those who had no evidence of inflammation [normal pouch] at baseline [T1], and who were not exposed to antibiotics. These selected, highly phenotyped patients were categorized into those who developed pouch inflammation within 1 year [Pre-Pouchitis group] and those who maintained a normal pouch for longer than 1 year [Normal Pouch-sustained group]. Although there were minor alterations of microbial groups in the Pre-Pouchitis patients, they had significantly decreased microbial diversity at T1 compared with patients who remained pouchitis-free. The possibility of subclinical inflammation in the Pre-Pouchitis group was ruled out since endoscopic and histological scores as well as faecal calprotectin levels were comparable between the two groups, indicating that the our findings were independent and, more importantly, that they preceded inflammation.

Of note, the decreased microbial diversity was detected up to 1 year before pouch inflammation and it was predictive of inflammation. An association between microbial diversity and gut inflammation had been previously demonstrated by others.^26,27 Kuhbacher et al.¹¹ showed that probiotics maintained remission in patients with pouchitis, and those authors attributed the protective effect to increasing gut microbial diversity. It is plausible to assume that low microbial diversity predisposes to microbial alterations that may result in inflammation. The observation by Hurst et al. that up to 50% of patients who underwent restorative proctocolectomy with ileal pouch-anal anastomosis surgery will experience pouchitis in the first year post surgery further supports this notion.⁷ Previous studies demonstrated that pouch microbiota is modified over time from an ileal to a colonic-like composition,^24,25 but the clinical significance of that observation is largely unknown. Having demonstrated that microbial diversity was significantly reduced during the first post-surgical year, we omitted that group of patients from our analysis, assuming that the microbiota profile during that time was not stable, possibly due to perioperative antibiotic therapy or due to the time it takes for the ileal pouch mucosa to be colonized by colon-like bacteria.^24,25 Interestingly, the patients who did not develop pouchitis during that year increased their microbial diversity over time [i.e. the four patients who were followed up for 3 years post surgery]. Given that decreased diversity is associated with inflammation, this may provide insight into the pathophysiological mechanism responsible for the development of pouchitis immediately after ileal pouch-anal anastomosis surgery.

Despite a prominent decrease in diversity, there were minor differences in microbial groups between the Pre-Pouchitis and the Normal Pouch-sustained patients. The presence of Ruminococous [Ruminococcaceae family] was significantly increased in patients who did not develop pouchitis. Ruminococcus degrades polysaccharides in the gut and provides important growth substrates which maintain intestinal mucosal integrity,^28,29 and it may also confer multiple health benefits including reversing infectious diarrhoea.^30,31Coprococcus and Lachnospira, both belonging to the Lachnospiraceae family, were also increased in the Normal Pouch-sustained patients compared with the Pre-Pouchitis patients. All three bacterial groups that were decreased in the Pre-Pouchitis group belong to the Clostridiales order, and are considered as being beneficial, butyrate-producing bacteria associated with gut health.^32,33 Reduced Bacteroides and multiple taxa of Lachnospiraceae and Ruminococcaceae families have been described in ileal Crohn’s disease.³⁴ We had recently shown that pouch inflammation is associated with a decrease in bacterial taxa, particularly from the families Lachnospiraceae, Ruminococcaceae and Bacteroidaceae.⁹ Moreover, Machiels et al.¹⁴ reported that the presence or absence of specific bacteria preceding pouch surgery can predict the future development of pouchitis. Thus, a combination of decreased diversity and absence of certain key beneficial bacterial groups may be a strong predictor and a driver of pouch inflammation.

The contribution of the microbiota to the pathogenesis of pouch inflammation is supported by previous evidence. Examples are: the lack of pouch inflammation before it is exposed to the faecal stream¹⁰; the response to antibiotic and probiotic therapy^11–13; and the correlation of certain microbial groups, such as the Fusobacteriaceae family, with disease activity and inflammatory markers.⁹ Other examples include the fact that the composition of faecal microbiota in patients with ulcerative colitis before pouch surgery, in particular the presence of R. gnavus, B. vulgatus and C. perfringens genera and the absence of Blautia and Roseburia genera, may be predictive of pouchitis development¹⁴ as well as the increased likelihood of pouchitis in hosts with a particular gene expression that have an abundance of certain bacteria, e.g. Escherichia.¹⁵

Intestinal inflammation is associated with alterations in the composition of the enteric microbial population and with a reduction in microbial diversity.^26,27 Despite an association of inflammation with loss [i.e. Faecalibacterium prausnitzii] or gain [i.e. the family Enterobacteriaceae] of certain bacterial groups, it is still not clear whether these alterations are the cause of inflammation or its outcome. It is speculated that the enteric microbiota can contribute to gut health through the production of beneficial metabolites, such as butyrate, that can enhance the epithelial barrier function, induce regulatory T cells and ameliorate the immune response.^35,36 However, increased microbial diversity by itself may also contribute to microbial stability, possibly by improving resilience against colonization by pathogenic bacteria, and it may help to maintain a healthy enteric microbial ecology.

The current long-term [40-month], prospective study that included meticulous clinical, endoscopic and microbial composition assessments demonstrated that decreased microbial diversity per se predicts pouch inflammation. Moreover, we focused on a highly phenotyped patient cohort in which microbial composition was not biased by previous exposure to antibiotic therapy or by inflammation. Our findings may provide an important tool for the clinician to predict which patients may be at risk to develop pouch inflammation after the first post-surgical year. In addition to suggesting a predictive tool, these findings have mechanistic implications pertaining to the pathogenesis of pouch inflammation whcih underscore the importance of microbial diversity. The clinical consequences may be the use of specific interventions in patients with a pouch, and possibly other inflammatory bowel diseases. Finally, although there was a significant difference in age between the two study groups, age is not a recognized risk factor for pouchitis and the age differences in our cohort did not seem to affect pouch diversity or composition [Supplementary figure 3].

Our study had some limitations; the patient cohort included is small. This is mainly due to the strict inclusion criteria and the need for longitudinal follow-up. A larger prospective study is required in order to substantiate the causative effect of reduced microbial diversity on the development of pouchitis as well as to identify more specific changes in bacterial composition. In addition, multiple factors that may affect enteric microbial composition and diversity were not completely accounted for, for instance diet that may affaect microbial composition as well as diversity.³⁷ Medications other than antibiotics, such as proton pump inhibitors³⁸ and other medications, may also alter the enteric microbiota and should be accounted for in larger prospective studies.

The findings of this study demonstrate that reduced pouch microbial diversity and possibly also reductions in specific bacterial groups, may predispose patients who underwent ileal pouch anal-anastomosis to develop pouchitis. More importantly, finding these changes in a ‘normal’ pouch may predict impending pouchitis. This may necessitate different preventive therapy, such as the use of probiotics or immunomodulators at an early stage. Recently, faecal microbial transplantation has demonstrated significant efficacy in increasing microbial diversity in patients with recurrent Clostridium difficile infection.³⁹ Potentially, this therapy may be tested in patients with a pouch and evidence for decreased microbial diversity, in order to reduce the considerable morbidity associated with pouchitis.

Funding

This study was supported by a grant from the Leona M. and Harry B. Helmsley Charitable Trust.

Conflict of Interest

The authors of this manuscript report no conflict of interest.

Author Contributions

ID and HT recruited patients for this study. NM, ID, LR and UG designed the study. NM and NAC collected patients’ data. NM and NAC drafted the manuscript and all authors reviewed it. ID approved final version. NM performed the microbiota analysis with the aid of LR and UG.

Acknowledgments

To Esther Eshkol for her assistance with the professional English editing of this manuscript.

References

Author notes